Aggregate Layout for FAS2552 AFF VDI

[peter65]

Hi -

recently racked a fas2552 with NO external storage all SSD system. Upgraded it to 8.3P1. This system is for our VDI environment. (2 nodes) single chassis). Setup up 10G NFS and will present datastores as NFS. I was just curious since there are 24 disks and 2 nodes do I:

a)create one big aggr for this environment and planned to present two datastores to vmware.  OR

b)create two aggrs (split the disks)  12 disk aggr for node a and 12 disk aggr for aggr2  then present one nfs datastore from each aggr / controller?

Also i may carve out 1 TB just to try smb shares for hyper v (very small environment but that would be a different SVM.

thoughts?

thanks

p.s Yes I did initialize the system after the 8.3p1 upgraded as it had 8.2.2 on it out of the box and I wanted to take advantage of the ADP (advanced drive partitioning) where by the disks show as "shared" for boot slicing.. Just trying to get all the efficiency out of it that i can.

[YIshikawa]

Plan a)
Pros:
- Easy to manage space. You can create bigger volume.
Cons:
- Load is redirected to one controller. You can not balance load between 2 controllers by serving aggregates on each node.
- need to consider some restrictions when assigning data partition.
https://library.netapp.com/ecmdocs/ECMP1636023/html/GUID-FFC4B324-157C-4A56-B01F-CF109A1F99AD.html
https://library.netapp.com/ecmdocs/ECMP1636022/html/GUID-CA94F1EB-218F-4126-8732-C1AD9DBE9337.html

Plan b)
Pros:
- You can balance load between controllers by serving aggregates on each node.
- You are able to manage restrinctions regarding core and reconstruction by assigning both root and data partitions to same node.
Cons:
- maximum capacity of single aggregate is smaller than plan a).

[thomasb82]

Hi Peter,

I was wondering if you have first results and can share them?

How many VDI VMs have you running, how is the Filer performing?

Many thanks!